Articles

Found 2 Documents
Search

PEMODELAN ARIMA DAN DETEKSI OUTLIER DATA CURAH HUJAN SEBAGAI EVALUASI SISTEM RADIO GELOMBANG MILIMETER

JUTI: Jurnal Ilmiah Teknologi Informasi Vol 7, No 3, Januari 2009
Publisher : Teknik Informatika, ITS Surabaya

Show Abstract | Original Source | Check in Google Scholar

Abstract

The purpose of this paper is to provide the results of Arima modeling and outlier detection in the rainfall data in Surabaya. This paper explained about the steps in the formation of rainfall models, especially Box-Jenkins procedure for Arima modeling and outlier detection. Early stages of modeling stasioneritas Arima is the identification of data, both in mean and variance. Stasioneritas evaluation data in the variance can be done with Box-Cox transformation. Meanwhile, in the mean stasioneritas can be done with the plot data and forms of ACF. Identification of ACF and PACF of the stationary data is used to determine the order of allegations Arima model. The next stage is to estimate the parameters and diagnostic checks to see the suitability model. Process diagnostics check conducted to evaluate whether the residual model is eligible berdistribusi white noise and normal. Ljung-Box Test is a test that can be used to validate the white noise condition, while the Kolmogorov-Smirnov Test is an evaluation test for normal distribution. Residual normality test results showed that the residual model of Arima not white noise, and indicates the existence of outlier in the data. Thus, the next step taken is outlier detection to eliminate outlier effects and increase the accuracy of predictions of the model Arima. Arima modeling implementation and outlier detection is done by using MINITAB package and MATLAB. The research shows that the modeling Arima and outlier detection can reduce the prediction error as measured by the criteria Mean Square Error (MSE). Quantitatively, the decline in the value of MSE by incorporating outlier detection is 23.7%, with an average decline 6.5%.

DESIGN AND IMPLEMENT ADAPTIVE NEURAL NETWORK SOFTWARE FOR ROUTING DATA PROCESS IN COMPUTER NETWORK

SAINTEKBU Vol 1 No 2 (2008)
Publisher : KH.A.Wahab Hasbullah University

Show Abstract | Original Source | Check in Google Scholar | Full PDF (120.19 KB)

Abstract

The data transmission in computer network is very important. Therefore, this issue always need a serious attention, especially in the middle and wide area networking which consist of many routers. This Adaptive Neural Networks software  for routing data in computer network (which is called JST Router later) is designed to solve a routing problem for choosing the best data routing path using Backpropagation Algorithm of Neural NetworkKeywords : routing table , routing algorithms, routing protocols , Artificial Neural Networks .